Erdős Problem #1190 — Put … where the maximum is taken over all disjoint systems for which n1>mn_1 > m.

About 46 years old · traced to

Put

εm=max⁡∑1/ni\varepsilon_m = \max \sum 1/n_i

where the maximum is taken over all disjoint systems for which n1>mn_1 > m. Determine or estimate εm\varepsilon_m as well as possible. I could not even decide whether εm→0\varepsilon_m \to 0 as m→∞m \to \infty.

References

Additional references

P. Erdős, A survey of problems in combinatorial number theory, Ann. Discrete Math. 6 (1980), 89-115.

Progress summary

Refreshed
Claimed solved

An AI-generated claim says the problem has its sharp answer, but no independent proof has yet verified it.

Erdős Problem #1190 asks for the asymptotic size of the largest reciprocal sum from a disjoint family of residue classes with moduli exceeding mm. Erdős credited Mirsky and Newman with the basic bound, while the sharp asymptotic is now claimed as a consequence of work on Problem #202.

Known results

  • Mirsky and Newman: psilonm<1psilon_m<1.
  • de la Bretèche, Ford, and Vandehey: L(m)−1+o(1)<psilonm<L(m)−3/2+o(1)L(m)^{-1+o(1)}<psilon_m<L(m)^{-\sqrt{3}/2+o(1)}, where L(m)=exp⁡ ⁣(0˘003log⁡mlog⁡log⁡m0˘003)L(m)=\exp\!\left(\u0003\sqrt{\log m\log\log m}\u0003\right).
  • Erdős reportedly did not know whether psilonm→0psilon_m\to0.

April 2026 claimed resolution

GPT-5.4 Pro is credited with implying psilonm=L(m)−1+o(1)psilon_m=L(m)^{-1+o(1)} from a claimed sharp result for Problem #202. The claim includes a proposed argument and Lean formalization, but no independent proof or published corroboration was found.

Current status (as of April 2026): The classical bounds are established, but the claimed sharp asymptotic psilonm=L(m)−1+o(1)psilon_m=L(m)^{-1+o(1)} remains unverified, so the problem is still open.

Sources

Solutions 0

No solutions have been posted yet.